Friday, March 2, 2018

Verschlimmbesserung

Verschlimmbesserung ("fair-shleam-besser-oonk") is German for "a supposed improvement that actually makes things worse". In English, we do not have any phrase which is nearly so concise or elegant, but I think Milton Friedman put it best:

One of the great mistakes is to judge policies and programs by their intentions rather than their results.

People ask me: why should I trust the free markets? In annoyingly Socratic fashion, I usually flip the question around and pose: why should I trust policymakers to appropriately manage the economy instead? If knowledge in a society is diffuse and decentralized (which is especially inherent to today's fast-paced constantly-online interconnected society) and thus needs to be aggregated to discover optimal solutions, what is the best system to do that? For politics, we can probably agree that that optimal solution is open democratic free elections with maximized voter participation. For economics, aren't free markets similar to free elections, except instead of the stakes being political decisions (legislation), we are faced with economic decisions (resource allocation) where "votes" are instead weighted by dollar instead of capita? Both are knowledge-aggregation machines, just with different intended purposes (legislation vs. resource allocation) and vote-weighting schema (dollars vs. capita). Just as how networks are valued by the number of nodes and connections (in fact, it may scale quadratically or even exponentially), the effectiveness of a knowledge-aggregation system depends the number of participants: Wikipedia (millions of mostly non-experts) proved far more successful than Encyclopedia Britannica (thousands of experts) despite the greater average expertise of the latter's contributors.

To be clear, I am not arguing that policymakers make sub-optimal decisions because they are malicious or because they are idiots (although we seem to be seeing a lot of both recently). The issue has nothing to do with whether policymakers are inherently good or evil, or even whether they are competent or not. Even if they were enlightened philosopher-kings advised only by top Harvard PhDs, the fact remains that a managed economy only aggregates the knowledge of a small subset of people and thus is are following the Encyclopedia Brittanica model rather than the Wikipedia model.

Furthermore, the type of knowledge they gain is frequently flawed. Policymakers by nature try to target certain indicators tmeasure the degree of success of their policies. However, when a measure becomes a target, it ceases to be a good measure. This is known as Campbell's Law or Goodhart's Law. One of my favorite examples is the use of test scores as a way to measure educational attainment:

achievement tests may well be valuable indicators of general school achievement under conditions of normal teaching aimed at general competence. But when test scores become the goal of the teaching process, they both lose their value as indicators of educational status and distort the educational process in undesirable ways.

Interestingly, the test itself isn't the problem. This problem arises no matter what type of test as long as the stakes are high enough (funding or admission is dependent on high scores), because the system adapts (teachers change their methods to "teach to the test" and students change their learning methods in order to maximize test scores). The system as a whole is no longer in the same state (aggregate behavior has changed) that it was before the measure became a target. The result? Test scores lose explanatory power and become worse measures of basically everything, even as they go up. If you improve your IQ score through intensive study, have you actually become more intelligent?  The map is not the territory.

Why does this happen? Due to emergence, complex systems are more than the sum of their parts. In fact, any effort by policymakers to measure macro-variables in a complex system tends to get Heisenbergian. This is even more true in social sciences where the state of the system is determined by the aggregate behavior of many individuals, each of whom can and will change their behavior in reaction to policymakers' public actions.

This is also why every economic crisis looks different from the last one (e.g. subprime mortgage lending caused the last recession while a tech bubble caused the one before that). This is because after every recession, policymakers try their very best to analyze what went wrong and usually discover that certain indicators would have been clear warning signals if only they had been measured. Thus armed with this knowledge, policymakers then try very hard to target the those exact indicators to prevent them from reaching those dangerous red flag levels. However, in the act of targeting those indicators, they have altered the deep parameters of the system (similar to the observer effect, where the act of observing a complex system can change the state of that system).

I hope it is clear why intervention in managed economies is generally sub-optimal. However, to be comprehensive, I believe that there are certain conditions in which it is in everyone's best interests that policymakers do intervene in the free markets (after all, I'm not an anarchist). These conditions can be grouped into: 1) externalities, 2) central banks setting monetary policy, and a bit further down in priority: 3) counter-cyclical fiscal policy as well as 4) wealth/income distributional issues (people born with different amount of dollars and thus "votes"). However, there are actually several ways to address these while respecting free-market mechanisms without resorting to creating a bloated bureaucracy with an alphabet soup of government agencies. If there's interest, I may do a follow-up post. Maybe even if there's no interest...

--- 

This post was inspired and prompted by "I, Pencil" as well as Friedrich Hayek's classic essay "The Use of Knowledge in Society".